以RFID物聯網與智慧型代理人為基礎之即時企業研究

Abstract

面對客戶要求更多的客制化產品以大量生產的價格交付，企業必須重新設計其現有的營運模式成為即時企業的運作模式來達成到靈活的生產與迅速回應不斷變化的客戶需求。為了有效地管理即時企業，企業必須開發能即時管理動態企業運作流程與處理例外事件的資訊系統。無線射頻辨識技術(RFID)和物聯網的資訊架構因可協助企業製造客製化產品以及以有效的追蹤追溯跨供應鏈的零件和產品，被認為是管理產品客製化最有效的解決方案之一。另一方面，智慧型代理人技術能處理即時企業的動態系統並在結合RFID與企業物流上發揮關鍵作用，以更佳的管理即時企業。然而，RFID物聯網和智慧型代理人是兩個複雜的研究領域，因此在這兩種技術的基礎上建立一個即時企業是許多企業面臨的重大挑戰。該問題顯示對於即時企業資訊系統目前還缺乏一連貫的系統塑模和設計方法。為解決此問題，本研究的主要目標在發展出一套一致且完整的系統塑模和設計方法來協助建構構一個基於RFID物聯網與智慧型代理人的即時企業。 在建構基於RFID物聯網與智慧型代理人的即時企業有下列幾個具有挑戰性的議題：（1）如何分析和開發基於RFID物聯網的流程與系統模型，（2）如何設計高階的代理人社會模型並將模型運用在RFID物聯網的系統流程管控上，使智慧型代理人的控制範圍以及其互動模式可以被定義;及（3）針對高階的代理人社會模型中所定義的智慧型代理人，如何進一步的發展其認知與內部控制模型，使他們能夠被實際運用並有效管理基於RFID物聯網的即時企業運作。 因此本研究將針對上述議題提出一建構基於RFID物聯網與智慧型代理人之即時企業的系統塑模和設計方法論， 該方法論可分為三個階段: （1）發展RFID物聯網的流程分析方法以及系統塑模方法與演算法，使基於RFID物聯網的流程與系統模型可被驗證;（2）發展代理人分析與設計的方法並以該方法為基礎在RFID物聯網系統模型上建構高階的代理人社會模型，（3）以第二階段發展的高階的代理人社會模型為基礎，進一步發展智慧型代理人的認知與內部控制模型，模型的建構方法基於BDI（信念，欲望，意圖）的推理模式以及代理人導向之多層次狀態反饋派翠網路(Petri net)。最後在系統實作面，第三階段的研究也提出一個系統實作的參考架構。 本論文的主要貢獻在提出一建構基於RFID物聯網與智慧型代理人之即時企業的系統塑模和設計方法論。為了驗證本研究方法論在實務上的可行性，本研究開發包含基於智慧型代理人技術的物聯網智慧節點與即時企業應用軟體之雛型系統來證明方法論的可行性，並作為開發即時企業系統的參考模式。最後透過雛型系統的實驗結果也證明，相對於目前的技術，本研究所提出基於RFID物聯網與智慧型代理人之即時企業系統確實能更有效的處理動態的企業運作，提升製造與物流自動化，以及降低在製品存貨與勞動成本。

Faced with customers demanding more customized products delivered at mass production prices, firms must reengineer their current business practices to a real-time enterprise (RTE) operational model to achieve flexibility in production and responsiveness to changing customer demands. To manage RTE effectively, firms must develop information system capable of managing dynamic business processes and operations in real time and taking immediate action to resolve disruptions. Radio Frequency Identification (RFID) technology and the Internet of Things (IoT) infrastructure are considered one of the most promising solutions for managing product customization as they enable firms to manufacture variants of product at a batch size of one and to effectively track and trace components and products across supply chain(s). On the other hand, intelligent agent based technologies have the capability to deal with the dynamic system of RTE and play a key role in integrating RFID with business processes to better manage an RTE. However, IoT/RFID and intelligent agents are two complex research fields and to build a RTE based on the two technologies is a major challenge facing many enterprises today. That problem indicates that there is a lack of coherent system modeling and design methodology for RTE systems. Towards the solution of this problem, the main objective of this research is focus on developing a unified yet comprehensive system modeling and design methodology to build a RTE based on the IoT/RFID and intelligent agents. Several challenging problems associated with the modeling and design of IoT/RFID and agent based RTE are: (1) how to analyze and develop IoT/RFID-driven processes and system model(s); (2) how to design and apply high-level agent social model(s) on the IoT/RFID system model so that controlling scopes of agents and agent-interaction schemes can be defined for IoT/RFID process management; and (3) how to develop internal cognition and control models of agents defined in the high-level agent social model so that they can be implemented and employed to effectively manage IoT/RFID-driven operations of a RTE. The proposed methodology in this thesis addresses the system modeling and design challenges present in building IoT/RFID and agent based RTE in a top-down approach: (1) Developing IoT/RFID process analysis and system modeling schemes and algorithms so that IoT/RFID-driven process and system models can be constructed and verified; (2) Developing agentification methods to build a high-level agent social model based on the IoT/RFID system model; and (3) Based on the high-level agent social model as a guide, devising agent’s internal cognition and control models using BDI (belief-desire-intention) reasoning model and agent-oriented hierarchical state-feedback Petri net approach. An implementation framework is also proposed to address the implementation issue. The key contribution of this thesis is to present a unified modeling and design methodology for building IoT/RFID and agent based RTE systems. To demonstrate the methodology’s real-world applicability, prototypical systems consisting of intelligent-agent based IoT smart end-point devices and real-time enterprise applications are developed to substantiate the applicability of the proposed methodology and to serve as a reference model for developing RTE systems. Results of prototype experiments also demonstrate that the proposed IoT/RFID and agent based RTE system is more capable than most current industrial practices in managing dynamic enterprise operations, automating manufacturing and logistic processes, and reducing work-in-process inventory and labor cost.